Skip to main content
Deep AI Detector
Log in Sign up free
API changelog · Preview

Deep AI Detector API changelog

Roadmap to v1.0 launch. The v1 major will be frozen against breaking changes for at least 6 months before deprecation; additive changes (new fields, new endpoints, new event types) will ship as minor versions. Make sure your AI detection client tolerates unknown fields when you wire it up after launch.

How the AI detection API changelog is organized

Every entry below is a release of the Deep AI Detector REST API — the same advanced AI content detection model that runs in the web dashboard, exposed over JSON for product teams, academic integrity platforms, and trust-and-safety pipelines. We treat backwards compatibility as a contract: clients that integrated against v1.0 will continue to work against v1.4 without code changes. The only obligation on your side is to tolerate unknown fields in response bodies, because additive minors are how we ship new capabilities without forcing a major bump.

The roadmap above is published in advance so engineering teams evaluating Deep AI Detector can plan integrations against a known target. Once the public API launches, every release will also include a deprecation notice for anything we plan to remove in the next major and a migration guide for the changes that matter.

Compatibility policy

  • Major (v2 etc): incompatible changes to existing endpoints. Always shipped with at least 6 months of overlap with the prior major. We email every API key holder on the deprecation announcement.
  • Minor (1.x): additive changes — new endpoints, new optional request parameters, new fields on responses, new event types. Safe to upgrade automatically.
  • Patch: bug fixes, performance, security. No observable contract change.

v1.4.0

minor Coming soon

API key management endpoints + readiness probe.

  • added GET / POST / DELETE `/v1/account/api-keys` for programmatic key lifecycle.
  • added GET `/v1/account` returns account info, billing mode, balance, retention preference.
  • added GET `/v1/health/deep` — per-component readiness (database, cache, ML inference, translation, billing, email).
  • added Webhook `balance.low` event when PAYG / API-only balance falls below $5.
  • added X-RateLimit-Limit / X-RateLimit-Remaining / X-RateLimit-Reset headers returned on every response.

v1.3.0

minor Coming soon

Deep-scan strategies + per-paragraph uncertainty band.

  • added `mc_dropout_band` and `combo` strategies on `/v1/detect/deep`. Returns calibrated uncertainty.
  • added `ai_prob_std` and `confidence_band` fields on deep response.
  • changed Deep response `band` now comes from the full aggregator for consistency with `/v1/detect`.

v1.2.0

minor Coming soon

Pricing quote on every detection.

  • added `pricing_quote` field on every `/v1/detect` and `/v1/detect/deep` response showing both billing modes.
  • added Per-API-key billing mode selection (per_word vs per_detection). Switch any time in the dashboard.
  • added Webhook `subscription.activated` / `subscription.canceled` / `subscription.past_due` events.

v1.1.0

minor Coming soon

Plagiarism endpoint goes GA.

  • added `/v1/plagiarism` — multi-source scan (open web + 5 academic sources + Wikipedia).
  • added `/v1/plagiarism/scans` history endpoints.
  • added Webhook `plagiarism.completed` event.

v1.0.0

major Coming soon

v1 public API GA. Compatibility guaranteed through at least 2026-09.

  • added `/v1/detect` — advanced AI detection model. Industry-leading accuracy trained on 2B+ samples.
  • added `/v1/detect/deep` — per-paragraph deep scan with multiple analysis strategies (paragraph_vote, sliding_window).
  • added `/v1/usage` — billing-cycle usage + per-day breakdown.
  • added `/v1/health` — liveness probe (no auth).
  • added Webhook `detection.completed` event with HMAC-SHA256 signing.
  • added Official @deepaidetector/client (npm) and deepaidetector (pypi) SDKs.

Subscribe to API release notes

RSS: /changelog.rss. Or email [email protected] to be added to the deprecation announcement list (sent only when a major bumps).

API changelog FAQ

How often does the Deep AI Detector AI detection API ship?

Roughly every 3–6 weeks for minor versions, with patch releases (bug fixes, performance, security) shipping continuously. Majors only happen when a breaking change is unavoidable — we expect v2 to be at least 18 months after v1.0 GA.

What counts as a breaking change?

Anything that could make a previously-working client stop working: removing a field from a response, renaming an endpoint, tightening a validation rule, changing the meaning of a score, or removing a webhook event type. New fields, new endpoints, new optional request parameters, and new event types are additive — they ship as minors.

Can I depend on the AI detection score format being stable?

Yes. The score (0–1 float) and band (categorical string) contract is part of v1 and will not change inside the major. If we ever recalibrate the model we will publish a calibration note in the changelog so you can decide whether to re-baseline downstream thresholds.

Where do I report issues or request new endpoints?

Email [email protected] with a reproducible example, or open an issue on the SDK repositories. We triage public issues weekly and respond to security reports within one business day.